research open access farm types and their economic ... · identification and characterization of...

Goswami et al. Agricultural and Food Economics 2014, 2:5http://www.agrifoodecon.com/content/2/1/5

RESEARCH Open Access

Farm types and their economic characterization incomplex agro-ecosystems for informed extensionintervention: study from coastal West Bengal, IndiaRupak Goswami1*, Soumitra Chatterjee2 and Binoy Prasad1

* Correspondence: [email protected] Rural Development andManagement Faculty Centre,Ramakrishna Mission VivekanandaUniversity, Narendrapur,Kolkata-700103, INDIAFull list of author information isavailable at the end of the article

©Am

Abstract

Identification and characterization of farming systems simplify huge diversity offarm types in complex agro-ecosystems, which is of critical importance for precisetechnological intervention and informed policy support. Multivariate statisticaltechniques like Principal Component Analysis (PCA) and Cluster Analysis (CA) maybe used for a wide variety of situations associated with farm typology delineation.The present study conducted in coastal saline India demonstratively establishedthe usefulness of such methodology in identification of predominant farm typesand their characterization. Data collected from 144 farm households throughquestionnaire survey could identify four predominant farm types with differentialincome sources and resource-base. The methodological perspective employed inthe study may be used as a decision support tool by extension agencies. On otherhand, a differentiated, holistic and broad-based extension intervention with suitableinstitutional arrangement will be needed to address the need of these identified farmtypes. This will lead to a reduced transaction cost of the agricultural research andextension systems in diverse ecosystems in India and many similar situations in thedeveloping countries.

Keywords: Cluster analysis; Economic characterization; Extension support; Farmheterogeneity; Farm typology; New Institutional Economics; Principal componentanalysis

BackgroundAdoption of new technologies in agriculture is of central interest to both academicians

and policy makers since this is directly related to the efficiency of an agricultural

research and extension system (Bozeman 2000). Higher adoption ensures higher return

on investment in research and development and creates a cycle of production-led

impact on economy and rural livelihoods. Unfortunately, there are numerous examples

of technologies with great potential that have not been accepted by farmers, especially

the smallholders of the developing countries. Quite often, these technologies do not

fit well into heterogeneous smallholder systems, which need specific technological

solutions. Such inherent variability often influences farmers’ response to various

technologies that aim at improving farm productivity and natural resource management

(Lal et al. 2001; Emtage and Suh 2005). Unfortunately, both in agricultural and social

sciences, complexity and diversity has been under-perceived and undervalued resulting in

2014 Goswami et al.; licensee Springer. This is an open access article distributed under the terms of the Creative Commonsttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in anyedium, provided the original work is properly cited.

mailto:[email protected]

mailto:[email protected]

Goswami et al. Agricultural and Food Economics 2014, 2:5 of 24http://www.agrifoodecon.com/content/2/1/5

their neglect, under-estimation and exclusion from government statistics and policy

framework (Chambers et al. 1989). The archetypal Green Revolution technologies and

‘transfer-of-technology’ paradigm has also historically failed to cater to the needs of these

diverse resource-poor agro-ecosystems in the developing countries (Pender and Hazell

2001). Extension offered blanket recommendation for wide geographical areas and was

largely used as a deterministic ‘dart gun’ (Roling 1988), i.e. ‘take the technology and

transfer it to farmers’. The heterogeneity of the farming systems for which different

technologies are needed, has been ruefully ignored.

Thus, study of farm typology is of practical interest for precise and effective technological

interventions. Farm typology study recognizes that farmers are not a monolithic group and

face differential constraints in their farming decisions depending on the resources available

to them and their lifestyle (Soule 2001). Ellis (1993) observes that small farmers are always

and everywhere typified by internal variations along many lines. Although every farm

and farmer is unique in nature, they can be clustered into roughly homogeneous

groups. Developing a typology constitutes an essential step in any realistic evaluation of

constraints and opportunities that farmers face and helps forwarding appropriate

technological solutions, policy interventions (Ganpat and Bekele 2001, Timothy 1994;

Vanclay 2005), and comprehensive environmental assessment (Andersen et al. 2009).

Moreover, typology studies are of paramount importance for understanding the fac-

tors that explain the adoption and/or rejection of new technologies (Kostrowicki

1977; Mahapatra and Mitchell 2001).

The heterogeneity of farming systems are created by a host of biophysical (e.g. climate,

soil fertility, slope etc.) and socio-economic (e.g. preferences, prices, production objectives

etc.) factors (Ojiem et al. 2006). Researchers have examined factors such as farm resources

such as cash and labor (Tittonell et al. 2007), infrastructure such as agency and markets

(Tittonell et al. 2006), management practices (Tittonell et al. 2005) and technological level

(Gómez-Limón et al. 2007). Few others have used string of factors together to explain the

heterogeneity of farming systems (Ojiem et al. 2006; Bidogeza et al. 2009; Guto et al.

2010; Tittonell et al. 2010; Rowe et al. 2006). The selection of factors that define farm

typology varies greatly from study to study and may be governed by the purpose of

research. For example, farm typologies were used to study appropriate fertilizer

application (Tittonell et al. 2006), resource use efficiency (Zingore 2007; Tittonell et al.

2007), water use efficiency (Senthilkumar et al. 2009), or overall classification of farm

types (Bidogeza et al. 2009), IPM (Leeson et al. 1999) and may sometimes be crop specific

in nature (Mwijage et al. 2009; Reckling 2011). Most of the farm typology study has

focused on socio-economic and agro-ecological factors for classification of farms.

Economic factors have been less used, especially in small-scale studies, for classifying

farms (USDA, ERS 2000; Briggeman et al. 2007; Andersen 2009). With economic

characterization of farms as the objective at hand, the present study assumes that

classification of farms based on economic returns from farm enterprises, along

with other related non-economic factors, will give more effective insights into the

farm type identification.

Indian farmers are heterogeneous in terms of agro-ecology and resource endowments

and the transfer of appropriate technology requires careful targeting. This problem is

exacerbated in complex-diverse-risk-prone coastal agro-ecosystems of the country

covering approximately 10 Mha cultivated area, which is characterized by a host of


bio-physical and socio-economic constraints coupled with diversity in livelihoods.

At present, agricultural research is based on primarily 15 agro-climatic zones and

127 sub-regions based on comprehensive research review of states (Khanna 1989).

Socio-economic and structural factors are not reflected in such classifications.

Economists classify farmers based on landholdings, which is not comprehensive

enough for classifying complex farming systems. Indian Council of Agricultural Research

(ICAR) used several factors for classification of farming systems, but the bases were more

discussion-based with limited statistical applications, perhaps due to functional purpose

of the classification (Sanjeev et al. 2010).

At the micro level, classification of farmers might be of practical use for localized

technological solutions and extension support. During the height of farming system

research and extension paradigm, conceptualization of recommendation domain

wanted to address this issue. A recommendation domain is a group of farmers whose

circumstances are similar enough that they are eligible for the same recommendation

(Harrington and Trip 1984). This led to informed decisions in part of technology

managers and higher rate of technology integration in smallholder systems. Classification

of farming situation by the farmers has also been suggested by some authors (Conroy and

Sutherland 2004; Goswami et al. 2012). However, these were more project-based and were

hardly mainstreamed in the national research and extension systems in the developing

countries (Frankenberger et al. 1989), India being no exception. Participatory research is

not mainstreamed in National Agricultural Research System in India and state-owned

monolithic extension system is not prepared to deal with the need of small farms in

diverse ecosystems (Glendenning et al. 2010). This leads to poor adoption of technology

and large yield gap (Aggarwal et al. 2008). Extension support and policy intervention also

suffers due to lack of informed decision by public extension. Although there has been

experimentation with reorganized and decentralized systems of technology assessment

and refinement and revitalized public extension systems, development of sound analytical

tools for targeting extension has remained undermined till date. A sound methodology

for profiling farm typologies will help in rapid transfer of appropriate technology, precise

extension support and development of policy environment sensitive to immense diversity

of smallholder farms in coastal saline India.

Macro-economic perspective on this systematic problem leads us to the examination

of efficiency in agricultural research and extension system. Ruttan and Hayami (1984)

cites the shortcomings of ‘induced innovation’ model (Hayami and Ruttan 1981) that

considers factor prices and user demand to induce the development of appropriate

technology and the widely cited ‘diffusion of innovations’ theory (Rogers 1962, 1995)

that assumes diffusion to have taken place in an institutional vacuum. Unlike the neoclas-

sical theory in economics, New Institutional Economics recognized the importance

of institutions in explaining uneven performance of economies (North 1997) and

suggested that this type of analysis can lead to the necessary institutional developments

vis-à-vis technical developments generated by R&D systems (Kydd 2002). Within the

extension science literature, responses could be traced in the form of Agricultural Know-

ledge and Information System (AKIS) concept (Engel 1995) that sees agricultural

innovation as a function of multi-stakeholder process. Roling (1988) departed from

Rogers’ theory towards a precise technological targeting and drew on the marketing

research tradition (Kotler 1986) for the same.


Increasing focus on the role of institutions in differential economic growth has now

added new dimension to the study of technology transfer and appropriate technology

(Hall and Yoganand 2004; Radulovic 2005). New Institutional Economics (NIE)

that makes explicit considerations of institutional arrangement and institutional

environment (North 1997; Williamson 1990), are now frequently drawn on in the

literature on efficient extension mechanisms (that reduces transaction cost and

enhances economic efficiency) (Birner and Anderson 2007; Kherallah and Kirsten 2002).

However, the application of NIE was mostly focused on the budgetary constraints and

institutional evolution, markets, and institutional development (Pal et al. 2003).

Methodology for identification of farm types for precise technological intervention

and policy support, if mainstreamed, may be beneficial to the institutional arrangement

that reduces transaction cost in smallholder agricultural production by providing

appropriate technology to the smallholders rapidly. This is even more important in

the regime of open economy where smallholders will have to be served with less

transaction cost for long-term economic and environmental sustainability.

The present study was undertaken to identify the predominant farm types in coastal

agro-ecosystem of India and to characterize them by some important socio-economic

indicators. The article demonstrates the methodology of farm typology study when

farming systems are heterogeneous and in need of appropriate technology for agricultural

sustainability.

MethodsArea of the study

West Bengal is an eastern state of India and one of the major players in agricultural

production. It has a high population density (976 per sq km) and varied farming

systems. Agriculture is the mainstay of rural livelihoods and nearly 90% of the cultivators

are small and marginal farmers jointly holding 84% of the State’s agricultural lands.

Another 3 million landless families have earned the right to cultivate and grow crops on

their own land after enactment of Operation Barga system (land reform movement) by

the State Government (GoWB 2005). The high population density has resulted in un-

abated fragmentation of cultivable land and insufficient farm size to sustain liveli-

hoods. Moreover, increase in the price of agricultural inputs, uncertain price of

perishable agricultural produce, inadequate market infrastructure, distress sale of

produce by small and marginal farmers are some of the problems that pose serious

challenges to sustainability of farm sector in the state. The socio-economic condi-

tion of the farming community is declining in the absence of appropriate technical, so-

cial, financial and market interventions (NABCOM 2009).

Based on climate, soil and physiography, there are six agro-climatic Zones for West

Bengal. Among these, South 24 Parganas district comes under Coastal Saline Zone

(Figure 1) (Gajbhiye and Mandal 2008), one of the biggest Zones with high cropping

intensity (143%) i.e. the ratio of gross and net cultivated area in percentage terms, and

covering part or whole of 6 districts of West Bengal. Climate is tropical moist sub-humid

with 1796.2 mm rainfall, and wide range of air temperature (maximum 35.0°C, minimum

15.6°C). Rice is the main crop grown over different land terrains and seasons. Aus (spring

paddy), sesame and green gram in pre-kharif (early wet season), jute and aman rice in

Figure 1 Study locations in South 24 Parganas district of West Bengal, India (shown by red circles).


kharif (wet season) and wheat, different oilseeds and pulses, and potato in rabi (winter

season) are important crops. South 24 Parganas (22°32′N to 22.53°N & 88°20′E to 88.33°E)

has an area of 8165.05 km2 with population of 81,53,176 (sixth highest in India), of which

74.39% stay in rural areas. Percentage of households below the poverty line is 37.21, much

higher than the state and country average (26% and 29%). The district is one of the poor

districts of West Bengal having large number of resource poor farmers. Natural resource is

fragile and highly prone to degradation.

Joynagar II, Baruipur, Basanti and Gosaba, the sampled Community Development

Blocks, are distributed across the middle and eastern part of the district (Figure 1).

Percentage of cropped and irrigated area of the district was 393.47 thousand ha and

98.31 thousand ha in 2006-07. Net cropped area of the district is 3,72,290 hectares,

decreasing slowly due to the conversion of agricultural land for other purposes.

Production of food grain, pulse and oilseed, vegetables and fruits has registered a

growth in recent years (GoWB 2005).

The district is characterized by complex, diverse and risk-prone ecosystem, which is

perturbed by extreme climatic events, water logging in low lands for considerable part

of the year and soil salinity in winter and summer months resulting in dearth of

appropriate technology for the district. Population density has risen rapidly over


the last decade, with 819 people per km2 arable land (2011 Census) and at a

growth rate of 18.05% between 2001 and 2011 (GoWB 2005). Sustainable use of natural

resources in the face of high population density and ecosystem fragility is critically

important for the region. Overexploitation of natural resources is overwhelming in

this district. Farming has diversified greatly over recent years (Ghosh and Kuri 2005) and

technological intervention with appropriate technology has become critically challenging

to the extension system. That is why classification of farming systems is a pragmatic

starting point for facilitating appropriate technology in this region.

Samplinga

Following the guidelines of Indian Agricultural Statistical Research Institute, data collection

was organized using a multi-stage random sample survey method (Sukhatme et al. 1984).

Four blocks (namely Joynagar II, Baruipur, Basanti and Gosaba) were randomly selected

from 29 blocks of South 24 Parganas district. Then, three villages from each of the block

and 12 farmers from each of the village were selected randomly. Care was taken that four

farmers each were selected from marginal-small (2.50-7.50 bigha), medium (7.50-12.00

bigha) and large (>12.00 bigha) categories (1 bigha = 0.13 ha). A quick survey of these 144

households was conducted with a focus on socio-economic information and income of the

farms from different farm enterprises. This data was used for identification of predominant

farm types for the region. After the identification of predominant farm types, half of the

members from each of the identified farm types were selected randomly and were

interviewed in details for eliciting additional socio-economic and farm management

related information (to be used for characterization). Thus, a total of 72 farmers

were interviewed in details for characterization of farm types. This helped us to

avoid detailed and time-consuming interviewing with all 144 respondents.

Data

Instead of conducting a conventional way of identifying and characterizing farming

system based on the source of the highest gross income received (from crops, vegetables,

fruits and orchard, livestock, dairy, fishery, and poultry), the study employed multivariate

statistics for identification of farm types and their characterization. Data were collected in

two phases as mentioned in Sampling section.

Based on an extensive literature survey (Dixon et al. 2001; Fliegel et al. 1968;

Mohammad 1978; Singh et al. 1993), initial case explorations in the field and expert

counsel, a structured interview schedule was developed for the study. Apart from socio-

economic parameters of the households, farm size, infrastructural facilities, information

on cost of cultivation and yield of different crops, and price received by the farmers were

included in the data collection instrument. The draft interview schedule was then pre-

tested on 20 non-sampled respondents to incorporate necessary modifications. Descrip-

tion and measurement of some of the variables used in the study is given in Table 1.

Data were collected during May-June 2012 through personal interview method from

144 sampled households for quick survey in the first phase followed by a detailed

survey on 72 respondents in the second phase. The dataset was prepared and analyzed

using SPSS 17 (SPSS Version 17, SPSS Inc., Chicago, Illinois) and XLSTAT 2012 software

(Version 2012, Addinsoft, Inc., Brooklyn, NY, USA).

Table 1 Selected variables used in farm type identification and characterization

Variable Description

Education Index Weighted average of formal education received by the household members(illiterate = 0, primary = 1, middle = 2, secondary = 3, higher secondary = 4,graduate = 5 and post-graduate = 6).

Crop Diversification Measured by Herfindahl Index; index varies from 0 to 1, ‘1’ - completespecialization and ‘0’ - perfect diversification (Malik and Singh 2002)

System Cost of Cultivation Cost of all inputs used (in Indian Rupees), for all enterprises of the farmingsystem including the cost of family labor

System Gross Return Total revenue earned (in Indian Rupees) from all the components of thefarming system

System Net Return System Gross Return – System Cost of Cultivation

Cost Benefit Ratio Gross return from sale of output/Total cost of input used


Statistical analysis

Both univariate descriptive statistics and multivariate statistical techniques were

employed for the analysis of data. Multivariate statistical techniques have been widely used

for farm typology and characterization study (Kobrich et al. 2003; Andersen et al. 2009;

Guto et al. 2010). These statistical techniques allow us creating farm typologies, particularly

when an in-depth database is available. The combination of principal component analysis

for necessary reduction of the number of variables followed by cluster analysis to identify

typical farm households was used. Similar examples may be drawn from Gebauer (1987)

in Germany that employs cluster analysis; Jolly (1988) in Senegal that employs

canonical-discriminant analysis for identification of recommendation domains; Orr

and Jere (1999) in Malawi employing cluster analysis followed by profiling through

discriminant analysis; Hardiman et al. (1990) in China using cluster analysis;

Köbrich et al. (2003) in Chile and Pakistan and Usai et al. (2006) in Sardinia, both

using principal component analysis and successive cluster analysis.

For the present study, farm household typologies were identified by using two

sequential multivariate statistical techniques: principal component analysis (PCA) and

cluster analysis (CA) (Ding and He 2004). PCA condenses the information from the

interdependent variables to a smaller set of factors (Jolliffe 2002; Abdi 2007). We used

both socio-economic variables and income from different crop enterprises in the PCA.

Since the impact of off-farm income on technology adoption is well-reported, the

present study also used off-farm income as a factor for classification (Savadogo et al.

1998; Nehring et al. 2005; Fernandez-Cornejo et al. 2005).

Factors were identified using orthogonal rotation (varimax method as in Kaiser 1970;

Gorsuch 1983) so that a smaller number of highly-correlated variables might be put

under each factor and interpretation becomes easier (Field 2005). In accordance

with Kaiser’s criterion, all factors exceeding an eigenvalue of one were retained

(Kaiser 1970). Kaiser’s criterion is accurate when the number of variables is less

than 30 (Field 2005), which was the case for our data set.

The sampled farms were clustered based on the five principal components identified

by PCA. Cluster analysis is a collective term covering a wide variety of techniques for

defining natural groups or clusters in data sets (Anderberg 1973). These groups are

relatively homogeneous within themselves and heterogeneous between each other

based on a defined set of variables. In a word, it is the art of finding groups in data

(Kaufman and Rousseeuw 1990). Entities within a certain group or cluster are expected


to be very similar to each other, and entities belonging to different classes are expected

to be dissimilar (Bidogeza et al. 2009).

To determine the number of clusters two steps were followed for clustering – the

hierarchical method and K-means clustering method. For hierarchical clustering,

Euclidian distance and Ward’s computation method was considered. The number of

clusters retained from Ward’s method (four in our study) was used as starting values in

the K-means method. Accordingly, the number of clusters that seemed most realistic

and meaningful was chosen for the final solution.

In addition to CA, a one-way analysis of variance test was performed. The test

allowed us to identify the differences in variance between clusters in terms of PC scores

(Field 2005). Thus, the variables that explained the largest differences between clusters

could be identified.

Results and discussionDescription of the farming systems

Four major types of farms, based on the source from which maximum gross income

was earned by the farmers, were identified. These were – rice-based farming system

(34 households), vegetable-based farming system (70 households), fishery-based

farming system (10 households) and farming system based on off-farm income

(20 households). The farm size wise distribution of these farm types in four study

Blocks has been reported in Figure 2.

A general observation revealed that large proportion of farmers followed rice-based

farming systems, perhaps to grow rice for self-consumption and not for sale. On the

other hand, medium farmers followed vegetable-based farming systems and marginal-small

farmers relied on off-farm income sources. This might be due to the pressing need for

income from their small holdings. Medium farmers could allocate enough land for

vegetables after growing rice for subsistence purposes. However, vegetable-based

Figure 2 Farmer category wise sub-farming systems in four study blocks of South 24 Parganasdistrict.


farming systems were found across farm sizes, perhaps for their high economic return.

Chili, tomato, brinjal, ladies finger, and cole crops were common in this part of the state

(Basu et al. 2009; Halder and Das 2012). Fishery-based farming systems were found

among medium and large farmers only, who could afford to spare land for water

bodies, in case of dug pond, and invest in commercial fishery. Spatially, most of

the vegetable-based systems were found in Joynagar-II Block followed by Baruipur

Block. In Basanti and Gosaba Block, all types of farm were found. A considerable

proportion of these farms earned large off-farm income.

Had we used the conventional method for identifying farm types, i.e. classification

based on farm size or income sources of the farm, the diversity of farming systems

would have been grossly underestimated. For example, it is difficult to understand

the importance of fishery (only 10 households) and animal husbandry (only four

households, not featured in Figure 2) in different farming systems if we go by the

numbers in Table 2 only.

In all the four study blocks, a large number of sub-farming systems were found.

These sub-farming systems were determined by unique combination of farm enterprises

(Table 2). The label of enterprises in a given sub-system was given according to their

contributions to the gross farm income. For example, ‘Rice + Jute + Vegetables’ sub-system

denoted that Rice contributed highest to the farm income, followed by Jute and

Vegetables. Number of sub-farming systems in Joynagar-II, Baruipur, Basanti and

Gosaba Blocks were 18, 16, 16 and 18 respectively. This huge diversity in sub-farming

systems (68 in no.) was inherent to the farms of this region and they ideally required

specific extension interventions. However, in practice, it was difficult to address the needs

of all these 68 sub-farming systems.

Since these sub-systems are huge in number and farm size or enterprise-based

classification is oversimplification of farm heterogeneity, we followed an alternative

methodology for identification of farm types and their characterization by employing a

combination of PCA and CA with the derived principal components. The results of these

multivariate analyses are described below.

The principal component analysis

Table 3 shows rotated factor (Varimax) matrix of independent variables with factor

loadings for each variable. The communality column shows the total amount of

variance of each variable retained in the factors. For the interpretation of the PCs,

variables with high factor loadings and high communality were considered from

the rotated factor matrix (Harris 2001). In total, 17 variables were included in

PCA, of which 5 principal components with eigenvalues greater than 1 were

retained for further analysis. These five PCs explained 73.46% of total variability in the

dataset. A closer look at each column of Table 3 helps us to define each component

according to the strongly associated variables. The first component explains 27.78%

variance and is correlated with the land holding, owned land, land fragment and income

from rice crop. Thus the component represents land entitlement and income from rice

crop. Principal component 2, 3, 4 and 5 explain 15.87%, 11.02%, 9.59% and 8.20% variance

respectively. That means the first two factors have more importance in explaining the

variance in dataset than the other three factors combined. PC 2 is correlated with crop

Table 2 Predominant sub-farm types under each farm types in the study blocks; sub-farm types are arranged according to the decreasing contribution of anenterprise towards gross farm income

Farmingsystems

Sub-farming systems

Joynagar-II Baruipur Basanti Gosaba

Rice-based Rice + Off-farm + Vegetables +Livestock + Pulses + Fishery +Fruit + Poultry

Rice + Vegetables + Off-farm +Livestock + Fishery + Poultry

Rice + Vegetables + Off-farm +Livestock + Fishery + Fruit +Goatery + Poultry

Rice + Vegetables + Off-farm + Livestock +Fishery + Goatery

Rice + Vegetables + Fishery +Livestock + Plantation + Fruit +Poultry

Rice + Vegetables + Fishery +Livestock + Goatery + Plantation +Fruit

Rice + Vegetables + Off-farm + Fishery +Livestock + Fruit

Rice + Poultry + Vegetables +Fishery + Plantation + Livestock

Rice + Vegetables + Off-farm +Fishery + Livestock + Fruit +Goatery

Rice + Fishery + Vegetables + Livestock +Goatery + Fruit + Poultry

Rice + Vegetables + Off-farm +Livestock + Fishery + Tree + Poultry

Rice + Vegetables + Off-farm +Fishery + Livestock + Fruit

Rice + Off-farm + Fishery + Livestock +Goatery + Poultry

Rice + Fruit + Off-farm + Vegetables +Livestock + Fishery + Plantation

Rice + Vegetables + Off-farm +Fishery + Livestock

Rice + Livestock + Vegetables + Fishery +Farm Machinery + Goatery + Poultry

Rice + Vegetables + Off-farm +Fruit + Livestock + Fishery

Vegetables-based Vegetables + Rice + Livestock +Pulses + Fishery + Goatery +Tree + Poultry

Vegetables + Rice + Off-farm +Fishery + Livestock + Oilseed

Vegetables + Rice + Fishery +Off-farm + Livestock + Fruit +Goatery + Poultry

Vegetables + Rice + Livestock + Fishery +Goatery + Fruit crop

Vegetables + Rice + Pulses +Fruit + Fishery + Livestock +Poultry

Vegetables + Rice + Fruit +Livestock + Fishery + Goatery

Vegetables + Rice + Off-farm +Fishery + Fruit + Livestock +Plantation + Goatery

Vegetables + Rice + Livestock + Fishery +Fruit + Goatery + Poultry

Vegetables + Rice + Fruit crop +Pulses + Livestock + Oilseed +Fishery + Plantation + Poultry

Vegetables + Rice + Fruit crop +Fishery + Livestock

Vegetables + Fishery + Rice +Livestock + Poultry + Fruit

Vegetables + Rice + Fishery + Fruit crop +Fishery + Livestock

Vegetables + Rice + Fruit +Fishery + Goatery + Pulses +livestock

Vegetables + Off-farm + Rice +Livestock + Goatery + Poultry

Vegetables + Rice + Fishery +Plantation + Goatery + Fruit +Poultry

Vegetables + Rice + Fishery + Livestock +Tree + Poultry

Vegetables + Rice + Fruit +Pulses + Fishery + Livestock +Oilseed + Plantation

Vegetables + Poultry + Fruit +Rice + Livestock + Fishery +Pulses

Vegetables + Rice + Off-farm +Fishery + Livestock + Poultry

Vegetables + Rice + Off-farm + Livestock +Fishery + Goatery

Gosw

amiet

al.Agriculturaland

FoodEconom

ics2014,2:5

Page10

of24

http://www.agrifoodecon.com

/content/2/1/5

Table 2 Predominant sub-farm types under each farm types in the study blocks; sub-farm types are ar nged according to the decreasing contribution of anenterprise towards gross farm income (Continued)

Vegetables + Rice + Pulses +Fruit + Fishery + Livestock +Tree + Poultry

Vegetables + Rice + Fishery +Goatery + livestock

Vegetables + Off-fa + Rice +Fishery + Livestock oultry

Vegetables + Rice + Fishery + livestock +Poultry + Fruit crop

Vegetables + Rice + Fruit +Plantation + Pulses + Fishery +Livestock

Vegetables + Rice + Fruit +Fishery + Livestock + Goatery

Vegetables + Rice + Oilseed +Goatery + Fruit + Fishery +Livestock + Poultry

Vegetables + Rice + Off-farm +Goatery + Fishery + Tree + Fruit +Poultry

Vegetables + Rice + Fruit +Fishery + Pulses + Plantation +Livestock

Vegetables + Rice + Fishery +Goatery + Pulses + Plantation

Vegetables + Rice + Livestock +Pulses + Fishery + Tree + Fruitcrop

Vegetables + Rice + Pulses +Fruit + Livestock + Plantation +Fishery + Oilseed

Vegetables + Rice + Livestock +Fishery + Tree + Goatery +Poultry

Vegetables + Fishery + Rice +Fruit + Pulses + Plantation +Livestock + Cotton

Vegetables + Rice + Off-farm +Goatery + Fishery + Livestock +Pulses + Plantation

Gosw

amiet

al.Agriculturaland

FoodEconom

ics2014,2:5

Page11

of24


/content/2/1/5

ra

rm+ P

Table 2 Predominant sub-farm types under each farm types in the study blocks; sub-farm types are arranged according to the decreasing contribution of anenterprise towards gross farm income (Continued)

Fishery-based Fishery + Vegetables + Cereal +Fruit + Livestock + Oilseed

Fishery + Cereal + Livestock +Goatery + Poultry

Fishery + Vegetables + Rice + Livestock +Goatery + Fruit + Poultry

Fishery + Cereal + Vegetables +Livestock + Poultry

Fishery + Rice + fruit crop + Livestock +Poultry

Fishery + Rice + Vegetables + Off-farm +Fruit crop + Livestock + Farm Machinery +Tree + Poultry

Off-farm based Off-farm + Cereal + VegetablesLivestock + Fruit + Poultry +Fishery + Tree

Off-farm + Cereal + Livestock +Fishery + Poultry

Off-farm + Rice + Livestock +Plantation + Tree + Fishery +Fruit

Off-farm + Rice + Fishery + Livestock +Poultry

Off-farm + Vegetables + Cereal +Livestock + Fishery + Tree + Poultry

Off-farm + Rice + Livestock + Fishery +Fruit + Poultry

Off-farm + Rice + Livestock + Poultry +Goatery + Fruit

Off-farm + Rice + Livestock + Vegetables +Fishery + Poultry + Fruit

Off-farm + Vegetables + Rice + Livestock +Poultry + Fishery + Tree

Off-farm + Rice + Livestock + Vegetables +Fishery + Poultry

Gosw

amiet

al.Agriculturaland

FoodEconom

ics2014,2:5

Page12

of24


/content/2/1/5

Table 3 Four principal components derived by principal components analysis withloadings for individual crop enterprises and percent cumulative variance explained

PC 1 PC 2 PC 3 PC 4 PC 5 Communality

Family Size 0.33 0.25 0.55 0.34 0.16 0.62

Land Holding 0.89 0.15 0.19 0.14 0.10 0.88

Own Land 0.69 0.16 0.14 0.19 0.17 0.60

Land Fragment 0.80 0.04 0.02 0.05 0.04 0.65

Crop diversification Index 0.22 −0.69 −0.07 0.26 0.05 0.61

Income from Rice 0.84 −0.12 0.09 −0.15 0.18 0.79

Income from Pulses 0.10 0.78 −0.16 0.34 −0.09 0.77

Income from Oilseed −0.03 0.59 0.19 −0.06 0.33 0.54

Income from Vegetables 0.46 0.25 −0.21 0.47 0.20 0.56

Income from Fruits 0.37 0.75 0.14 0.04 −0.17 0.74

Income from Tree −0.27 −0.10 −0.09 0.10 −0.47 0.53

Income from Plantation 0.28 0.63 −0.08 0.30 0.06 0.58

Income from Cattle 0.25 0.12 −0.07 0.10 −0.55 0.63

Income from Poultry 0.09 −0.04 0.92 0.00 0.01 0.86

Income from Fishery 0.41 −0.15 0.65 −0.02 −0.14 0.70

Income from Goatery 0.01 −0.15 −0.02 0.19 0.85 0.77

Off-farm income 0.07 −0.13 −0.19 −0.83 −0.08 0.79

Eigen values 4.72 2.36 1.36 1.29 1.05

Cumulative explained variance 27.78 43.66 54.68 63.26 73.46


diversification index and income from pulse, oilseed, fruits and plantation, representing

crop and income diversification. PC 3 is correlated with income from poultry, fishery and

family size, embodying the income from poultry and fishery enterprises driven by family

labor. PC 4 represents off-farm income and PC 5, correlated with income from goatery

and cattle, represents livestock enterprises. Thus the five principal components could

be names as ‘land ownership and income from staple food’ (PC 1), ‘farm diversification’

(PC 2), ‘income from poultry and fishery’ (PC 3), ‘off-farm income’ (PC 4), and ‘income

from livestock’ (PC 5).

The cluster analysis

The first five factors from the principal component analysis were used for hierarchical

clustering using Euclidean Distance as distance measure and Ward’s technique as

agglomerative clustering. The agglomeration schedule, resulting from this analysis,

illustrated the sequence of analysis and produced coefficients. The objective of generating

such schedule is to arrive at an appropriate number of clusters that best fit the data set. A

check of the agglomeration schedule and Scree Diagram suggested that the number of

clusters should be four. It was then checked that the number of retained clusters is realistic

with respect to the field observation in order to be accepted as a meaningful classification.

Thus, four clusters were found to be appropriate, as these seemed to be most representative

of farm households within the study blocks. The k-means clustering method was employed

with 4 numbers of clusters. The analysis produced final cluster centres (Table 4), and distri-

bution of farms within a cluster among study blocks (Table 5) and size of farms (Table 6).

Table 4 Final cluster centres for all four clusters identified through K-means clusteringmethod

Principal components Clusters

1 2 3 4

PC 1 (Land ownership and income from cereals) 0.84 0.23 −0.31 0.53

PC 2 (Farm diversification) −0.09 0.87 −0.17 0.53

PC 3 (Income from poultry and fishery) 3.58 0.10 −0.19 0.50

PC 4 (Off-farm income) −0.03 −0.20 −0.72 0.35

PC 5 (Income from livestock) −0.00 −0.50 −0.07 2.14


Final cluster centers interpret what is typical for a particular cluster. Cluster 1 is

characterized by relatively higher income from poultry and fishery enterprises, has

large holdings, and earns high from rice crop. Cluster 2 is characterized by higher

diversification of farm income and Cluster 3 is characterized by smaller farms with

higher off-farm income. Cluster 4 represents farms with higher income from livestock and

other diverse sources such as rice, vegetables, poultry and fishery (Table 4). In all clusters,

the contribution of several important crops and off-farm income could be observed.

Figures 3 & 4 (A, B) represents the observations and variables simultaneously in the

two dimensional space. The individual farms have been labeled with study blocks

(Figure 3A & B) and size category of farms (Figure 4A & B). The A series (3A & 4A)

depicts the distribution of observations in a space defined by two principal components,

after varimax rotation. The B series (3B & 4B) depicts the same along with the initial

variables plotted in the form of vectors. Farms clustered (circled) together represent a

farm type. A visual check suggests one group at the right and lower quadrant of the figure

(Figure 3A), another clustered at the left lower quadrant, which is closer to the axis. The

third one is also located at the left lower quadrant, but away from vertical axis. The

scattered farms of upper right and left quadrants form the fourth group. A check

of these Figures (3 & 4) in combination with Tables 5 and 6 helps us to understand

this clustering better.

Noticeably, type of farms is not specific to a study block. The labels ‘1’, ‘2’, ‘3’ and ‘4’

are scattered throughout the plot. However, in some places similar labels are clustered

signifying similar farms within a Block. The findings from Figure 4A are even more

interesting. Small and marginal farms labeled as ‘1’ are clustered together signifying

their similarity in terms of a host of factors. Large farms labeled as ‘2’ are also found

closely clustered. Instead, medium farms are scattered throughout the plot representing

Table 5 Distribution of farm types, identified by cluster analysis, in four study blocks ofSouth 24 Parganas districts

Farm types Total

Cluster 1 Cluster 2 Cluster 3 Cluster 4

Joynagar-II 0 (0.00, 0.00)* 18 (36.00, 50.00) 10 (14.29, 27.78) 8 (57.14, 22.22) 36

Baruipur 8 (100.0, 22.22) 8 (16.00, 22.22) 18 (25.71, 50.00) 2 (14.29, 5.56) 36

Basanti 0 (0.00, 0.00) 8 (16.00, 23.53) 22 (31.43, 64.71) 4 (28.57, 11.76) 34

Gosaba 0 (0.00, 0.00) 16 (32.00, 44.44) 20 (28.57, 55.56) 0 (0.00, 0.00) 36

Total 8 50 70 14 142

Likelihood ratio 46.13 Sig. = .000

*Figures in parentheses are percentages to column and row marginal respectively.

Table 6 Distribution of farm types, identified by cluster analysis, among three farmercategories in South 24 Parganas districts

Farm types Total

Cluster 1 Cluster 2 Cluster 3 Cluster 4

Marginal-small 0 (0.00, 0.00)* 4 (8.00, 8.33) 44 (82.86, 91.67) 0 (0.00, 0.00) 48

Medium 2 (25.00, 4.35) 26 (52.00, 56.52) 14 (20.00, 30.43) 4 (28.57, 8.70) 46

Large 6 (75.00, 12.5) 20 (40.00, 41.67) 12 (17.14, 25.00) 10 (71.43, 20.83) 48

Total 8 50 70 14 142

Likelihood ratio 67.089 Sig. = .000

*Figures in parentheses are percentages to column marginal and row marginal respectively.


higher instrinsic variability of medium farms. The nearness of blue dots (individual

farms) to the red dots (initial variables) shows the characteristics of those farms in

terms of the variables (represented by red dots). A close look of both 3A and 4A

provides better understanding of farm classification in the study areas.

It is observed that all the Cluster 1 farms (100%) are situated in Baruipur Bock

(Table 5). Cluster 2 farms are predominant in Joynagar II (36.00%) and Gosaba

(32.00%), Cluster 3 farms are distributed across the four study blocks and Cluster 4

farms are predominant (57.14%) in Joynagar II Block. The data suggest that Joynagar II

Block is predominated by diversified farms, Baruipur and Basanti Blocks by smaller

farms with higher off-farm income and Gosaba Block with diversified and smaller farms

with off-farm income. The distribution of clusters across the study blocks differed

significantly (Likelihood ratio sig. value = .00) indicating heterogeneous nature of farm

types in these blocks.

From Table 6, it may be observed that marginal-small farms are mostly Cluster 3

farms (91.67%), i.e. smaller farms with higher off-farm income. Medium farms are

Cluster 2 (56.52%) and Cluster 3 (30.43%) farms i.e. diversified farms and small farms

Figure 3 Distinct cluster of farms delineated by PCA axis and labeled by study blocks. A. Principalcomponent analysis observations plot diagram of farming systems based on PCA of variables after varimaxrotation. The numbers 1-4 on the plane indicate the community development blocks namely Joynagar II - 1,Baruipur - 2, Basanti - 3 and Gosaba - 4. The distance between the positions of the farms approximates thedissimilarity of their variables measured by Euclidean distance. Four circles/ovals represent four farm types.B. Ordination bi-plot diagram of farming system characterization based on PCA of variables. The numbers 1-4 onthe plane indicate the community development blocks namely Joynagar II - 1, Baruipur - 2, Basanti - 3 andGosaba - 4. Red dots represent variables used in PCA. The distance between the positions of the farmsapproximates the dissimilarity of their variables measured by Euclidean distance. Samples close to the originhave average values of a particular variable in a study sample.

Figure 4 Distinct cluster of farms delineated by PCA axis and labeled by size category of farms.A. Principal component analysis observations plot diagram of farming systems based on PCA of variablesafter varimax rotation. The numbers 1-3 on the plane indicate the different farm sizes – Marginal-Small - 1,Medium – 2, and Large - 3. The distance between the positions of the farms approximates the dissimilarity of theirvariables measured by Euclidean distance. Four circles/ovals represent four farm types. B. Ordination bi-plotdiagram of farming system characterization based on PCA of variables. The numbers 1-3 on the plane indicate thedifferent farm sizes – Marginal-Small - 1, Medium – 2, and Large - 3. Red dots represent variables used in PCA.The distance between the positions of the farms approximates the dissimilarity of their variables measured byEuclidean distance. Samples close to the origin have average values of a particular variable in a study sample.


with higher off-farm income. Noticeably, large farms are distributed across all clusters,

perhaps due to their higher flexibility in allocation of land resources among different

crops/enterprises. Alternatively, Cluster 1 farms are mostly (75.00%) large farms,

Cluster II farms are largely Medium (56.52%) and Large (41.67%) farms, Cluster 3 farms

are mostly Marginal-small farms (82.86%) and Cluster 4 farms are Large farms (71.43%).

The distribution of clusters across the farm size differed significantly (Likelihood ratio

sig. value = .00) indicating heterogeneous nature of farm types across farm sizes.

In some parts of West Bengal including South 24 Parganas district, departure from

traditional cropping pattern has been a trend for the last one and half decades

(Goswami 2007; Bhattacharyya 2008) for enhancing income from fragmented land

resources, which are challenged by abiotic stresses (e.g. salinity). The number of

smallholders has in fact increased in West Bengal (De 2003) and there is evidence of

increased crop diversity on fragmented lands in the state (De 2002; Bagchi et al. 2012).

In many places, farmers have changed land shapes to grow vegetables in low lands (Basu

et al. 2009). Introduction of farm-ponds and land-shaping has also resulted in high crop-

ping and income diversity among small farmers (Bahu and Mishra 2001). Off-farm

income has also become a burgeoning reality of rural India especially for the smallholders

and this has often become the largest source of rural farm income. This also contributes

to the investment in farming activity and asset creation (Mehta 2009).

To examine the authenticity of the cluster analysis, a one-way analysis of variance for

each principal component (equality of group means) was performed with the four

clusters (Table 7). The more distinct a PC value is among groups, the lower is the p-value.

Judging from the p-values, all the factors namely ‘land ownership and income from rice

crop’, ‘income from poultry and fishery’, ‘off-farm income’ and ‘income from livestock’

found to be significant in differentiating the clusters (p = 0.00 for all factors). However, fac-

tor 2 characterized by ‘farm diversification’ was not significantly differing across identified

clusters, although the p value was not far from significant (p = 0.09).

Table 7 One-way analysis of variance of factor scores across four identified clusters

Factors Cluster Error F Sig.

Mean square df Mean square df

PC 1 (land ownership and income from rice crop) 4.68 3 .92 138 5.10 0.00

PC 2 (farm diversification) 2.14 3 .98 138 2.20 0.09

PC 3 (income from poultry and fishery) 26.06 3 .46 138 57.25 0.00

PC 4 (off-farm income) 36.30 3 .23 138 156.12 0.00

PC 5 (income from livestock) 25.58 3 .47 138 54.95 0.00


Characterization of identified farm types

Table 8 helps us to understand the characteristics of the clusters in terms of

background variables and economic performance indicators of the farming systems.

Cluster I accounts for 5.63% of farm households. The cluster is comprised of households

having large land holding, large family size (representing joint extended families), relatively

higher family education, and relatively higher crop diversification (Col. 1, Table 8). The

households both lease out and lease in land, land are highly fragmented and situated at a

high distance from the dwellings. In terms of economic performance indicators, this cluster

is characterized by relatively high system gross return, higher cost of cultivation and system

net return and relatively higher cost-benefit ratio. These households, on an average,

secured high income from rice crop, poultry and vegetables followed by fruits and fishery.

They depended wholly on agriculture with no off-farm income. These farms may be

supported for technically sound intensification of agriculture with assured input and

advisory services. Since these groups pursue a capital intensive diversified farming, access

to credit is important for them. The farmers may be assisted in commercial production

techniques with provision for building requisite capacity in farm and non-farm based

entrepreneurship (Folmer et al. 2010). Export-oriented crops may be tried with these

farmers with suitable incentive (Singh 2002). Since these farms are capital and management

intensive, they are prone to become environmentally unsustainable in the long run. An

environmental monitoring mechanism is needed to be in place in addition to an

incentivized system in favour of sustainable farming (Dhiman and Thambi 2009).

Cluster II comprised of 35.21% of the farm households. The cluster members had

relatively smaller land, smaller family size, relatively poor family education, and high

crop diversification (Col 2, Table 8). These farms lease-in land and their fragmented

lands are relatively nearer to their settlement. This cluster is characterized by low

system gross return, low cost of cultivation, moderate system net return, but relatively

better cost-benefit ratio. These households secured large proportion of income

from vegetables, fishery, livestock and rice crop and may be grouped as diversified

farms. These farms are management intensive, presumably supported by human

labor, and pursue diversification in enterprises. Integrated farming with livestock

and fishery will help in higher food and livelihood security without undermining

farm resources (Prein 2002). This will require specialized extension support with a

farming system approach, and institutional convergence for sound planning (Nabi 2008).

Cash crops may be promoted for enhancing farm income with assured provision

for credit and insurance (Fafchamps 1992). On account of the large proportion of

such farms, collectivization might be a viable strategy for their market integration

(Hazell 2005).

Table 8 Characteristics of identified clusters of farm households and p-value of one wayanalysis of variance (equality of group mean) for variables used in characterization offarm types

Cluster 1 Cluster 2 Cluster 3 Cluster 4 P-value

(N = 4) (N = 25) (N = 35) (N = 7)

(1) (2) (3) (4)

Age (yrs) 50.50ab** 51.68a 46.03b 53.86a 0.00

Family size 9.50 6.72 5.60 7.29 0.00

Farm size (bigha)* 10.38a 7.71bc 5.28c 9.65ab 0.00

Own land (bigha) 6.88ab 6.55b 4.41b 9.65a 0.00

Leased in land (bigha) 3.50a 1.16b 0.87bc 0.00c 0.00

Leased out land (bigha) 0.28a 0.12b 0.07b 0.36a 0.02

Land fragment (no.) 4.25ab 3.96b 3.22c 4.43a 0.01

Distance from home 2.0a 1.08b 1.08b 1.28b 0.00

Education Index 2.27 1,92 1.92 1.97 0.20

Area of Rice (bigha) 8.00a 6.16ab 4.76b 7.79a 0.00

Area of Vegetables (bigha) 0.75b 1.40a 0.70b 1.84a 0.00

Area of oilseeds (bigha) 0.23b 0.86ab 0.09b 1.43a 0.00

Crop Diversification Index 0.29ab 0.33a 0.28b 0.28b 0.01

System Gross return (Rs.) 139640ab 110980b 66508c 149550a 0.00

System Cost of Cultivation (Rs.) 60391ab 47228b 30330c 64218a 0.00

System Net Return (Rs.) 79250a 63755a 34576b 79449a 0.00

Cost-Benefit Ratio 2.42 2.34 2.39 2.28 0.88

Income from Rice (Rs.) 37380a 23165b 20734b 34563a 0.00

Income from pulses (Rs.) 576c 3538ab 310bc 4071a 0.00

Income from Oilseed (Rs.) 1050b 296b 40b 2657a 0.00

Income from Vegetables (Rs.) 23610b 42148a 16817b 46093a 0.00

Income from Fruits (Rs.) 9700a 6584ab 29862 5185b 0.01

Income from plantation (Rs.) 1750b 1856b 332c 3614a 0.00

Income from Livestock (Rs.) 5600ab 6059a 4738b 6558a 0.01

Income from poultry (Rs.) 50125a 503b 701b 212b 0.00

Income from Fishery (Rs.) 8250b 18816a 4995 15800a 0.00

Income from goatery (Rs.) 1600b 650c 1145b 6485a 0.00

Off-farm Income (Rs.) 0c 800c 15171a 2857b 0.00

*1 bigha = 0.133 ha.**Difference in mean among clusters denoted by different letters as found from Duncan Multiple Range Test. Highestmean value is denoted by a, followed by b and c when significant mean difference was found between clusters.


Cluster III comprises of 49.30% of the sampled farms. This cluster is characterized by

small land holding, relatively poor family education, smaller family size and relatively

low crop diversification (Col. 3, Table 8). These farms have started shifting to off-farm

income to offset low income from farming. In terms of economic performance, this

cluster is characterized by low system gross return, low cost of cultivation or low

investment in farming, poor net return from farming, but higher benefit-cost ratio

(since they invest less). The farm households earn most of their income from rice crop,

vegetables and from off-farm sources. These are typical subsistence farms that have

started to shift from traditional farming to non-farming vocations, often in the form of

seasonal migration. Maintaining food security is a key policy goal for these farms,


which may be achieved through sustainable and risk-averting cultivation of food

grains (e.g. climate resilient agriculture) coupled with assured coverage under public

distribution system for food grains (Thomas 2006). This might be supplemented with a

comprehensive capacity building programme on non-farm enterprises, since these people

are poor in human resources and exploited in the labor markets in nearer towns

(Deb et al. 2002). Collectivization of small enterprises may follow this capacity

building programme for their market integration.

Cluster IV comprises of 9.86% of the farm households. The cluster members had

relatively larger land holding (owner cultivator), relatively larger family size, moderate

family education, and moderate crop diversification (Col. 4, Table 8). They do not

lease-in land, i.e. these farms represent owner cultivator who lease-out land occasionally.

Land is highly fragmented. These households secure high income from rice crop,

vegetables, and fishery and are characterized by high system gross return, high cost

of cultivation, high system net return and relatively low benefit-cost ratio. The

cluster members had little off-farm income. These farms are capital intensive, developed

to meet growing food demand in the nearest town markets. Unlike Cluster I, these farms

earn higher income from vegetables and fishery. These may be supported for intensive

farming with improved technology. Since they specialize is in food grain and vegetables,

assured marketing support and crop insurance are critical for these farms. Due to their

available family labors, sustainable agriculture and agro-based enterprises also seem to be

promising propositions (Bhattacharjee and Saravanan 2012).

This farm typology classification offers clear advantages over classifications based on

farm size or agro-ecological characteristics. The farm types delineated are manageable

in number and represent both socio-economic, resource ownership and management

orientation of the farms. Farm size-based classifications undermine the huge diversity

among size classes and agro-ecological classification and ignore socio-economic realities

of the farms. At the same time, the farm classes are based on sound statistical procedures

instead of size-based classifications and, hence, more acceptable to policymakers.

The four farm types differed among themselves in terms of all studied parameters

including a string of background variables, income sources and economic indicators,

showing the efficiency of the classification methodology employed. The farm types,

however, did not differ in terms of education index and cost-benefit ratio. Increasing

dependence on costly external inputs has perhaps rendered an increase in cost of

cultivation, relative to system net return, irrespective of farm types. Education was not

an important means to classify farms, as access and control over natural resource was

more important in differentiating farm types.

Conclusions and implicationsThe implications of the study may be illustrated in practical and academic terms. While

the practical focus is on institutional arrangement and policy interventions, the academic

focus has been on methodological insights gained and its possible upscaling.

The methodological implication and scope for modification and upscaling

The study departs from conventional methodology of economic characterization of

predominant farming systems (classification based on land holding or discussion


with stakeholders with little statistical rigor) in terms of at least two dimensions.

First, it uses multivariate statistics for identification of farm types, and second, it

uses a string of economic and non-economic variables for such classification. This

methodology has reduced the time of data collection significantly, as quick survey was

employed for farm type identification and only half of the households were interviewed in

detail for their economic characterization. In-depth and comprehensive classification of

farms was possible resulting in a manageable number of farm types (four) that could be

taken up for appropriate policy and technological intervention.

The methodology employed in this study can be modified under different situations

of farm typology study. The variables used for PCA may be different, based on the

nature of agro-ecosystem, nature of agriculture practiced, objective of classification

(nutrient management, irrigation intervention, credit support) etc. Variables may also vary

when the focus of farm characterization is different from economic characterization only

(e.g. these may be energy efficiency, ecological sustainability). The implication would also

reflect in the development of data collection instruments to be used in quick and detailed

survey. The process of variable selection may involve stakeholder participation, thus

promoting popular participation in the study. The multivariate approach (algorithm)

followed for the farm type delineation may also be used for the development of

software or web-based decision support system. Separate modules in the software

may be incorporated with flexible set of indicators under each of them. This may be used

by research and extension agencies after rigorous multi-locational field testing.

Proposed institutional environment for technology transfer

The results from the analysis suggested four major farm types in the study areas –

(i) capital-intensive large diversified farms with high economic efficiency, (ii)

management-intensive, medium, diversified farms with moderate economic efficiency, (iii)

small, subsistence farms with high off-farm income and high economic efficiency, and (iv)

large diversified owner-cultivated farms with relatively lower economic efficiency. Smaller

farms have inclined towards non-farm income more than the medium and large farms.

On the other hand, medium and large farms have diversified more than smaller farms

with a departure from grain-based systems to vegetable-based farming systems. This

asks for a holistic farm planning and extension intervention than confining efforts

to technology transfer alone. India, along with a large number of countries having

agrarian society, has entered the open economy regime and need to establish economic

efficiency for their smallholder system (Birner et al. 2009). The extension system

must precisely target agricultural inputs, advisory services, credit access and critical

information for identified farm. The selection of beneficiaries for many public ex-

tension programs may also be guided by such farm typology.

To respond to the need of such comprehensive farm types, institutional innovation

(development and/or convergence) often becomes necessary. In India, a nationwide

revitalized extension system has just started functioning. It explicitly encourages such

convergence in planning and execution of technology transfer programs. Linkage with

existing donor supported and Ministry funded efforts might also lead to improved

institutional environment. India is now experiencing a handful of flagship rural

development programs that builds on constitutional provisions for basic services to


its citizens and development of rural institutions (mostly in the form of rural

collectives), and these may be efficiently tapped for effective transfer of agricultural

technologies.

Extrapolation of the study outcome

The study does not necessarily claim to extrapolate the identified farm types to larger

context. But, since the study locations are representative of coastal saline zone of India,

this may be generalized for at least 10 Mha of cultivable lands with nearly 900 million

people operating under this system. However, it is the methodology that is subject to

generalization for even larger regions. India has almost 200 districts under complex

agro-ecological systems covering a population of nearly 400 million. Integration of the

methodology in the formal policy or in the form of web-based decision support

tool may reach these people effectively. Need not to say, the same may be explored

anywhere in the world using flexible and participatory indicator selection and their

subsequent validation.

EndnoteaLogistic support during sampling and data collection was received from All India

Coordinated Research Project on Integrated Farming System and academic research

programme of Ramakrishna Mission Vivekananda University.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsRG designed research, analysed and interpreted data and prepared the manuscript, SC conceptualized and designedthe research work, and BP collected data and drafted the manuscript with RG. All authors read and approved the finalmanuscript.

AcknowledgementLogistic supports received from the All India Coordinated Research Project on Integrated Farming System andAcademic Research Programme of Ramakrishna Mission Vivekananda University is duly acknowledged. Authors arealso thankful to all the farmers who assisted and participated in the survey.

Author details1Integrated Rural Development and Management Faculty Centre, Ramakrishna Mission Vivekananda University,Narendrapur, Kolkata-700103, INDIA. 2Department of Agricultural Economics, Bidhan Chandra Krishi Viswavidyalaya,Mohanpur, Nadia-741252, West Bengal, INDIA.

Received: 14 May 2013 Accepted: 7 April 2014

References
Abdi H (2007) Discriminant Correspondence Analysis. In: Salkind NJ (ed) Encyclopedia of Measurement and Statistics.
Sage, Thousand Oaks, CAAggarwal PK, Hebbar KB, Venugopalan MV, Rani S, Bala A, Biswal A, Wani SP (2008) Quantification of yield gaps in

rainfed rice, wheat, cotton and mustard in india, Global theme on agro-ecosystems report # 43. ICRISAT,Hyderabad

Anderberg MR (1973) Cluster Analysis for Applications. Academic, New YorkAndersen E (2009) Regional Typologies Of Farming Systems Contexts, seamless Integrated Project, Eu 6th Framework

Programme. Available via http://www.seamless-ip.org/Reports/Report_53_PD4.4.3.pdf. Accessed 10 Apr 2012Andersen E, Elbersen B, Godeschalk F, Verhoog D (2009) Farm management indicators and farm typologies as a basis

for assessments in a changing policy environment. J Env Manage 82:353–362Bagchi BD, Roy SB, Jaim WMH, Hossain M (2012) Diversity, spatial distribution, and the process of adoption of improved

rice varieties in West Bengal. In: Hossain M, Jaim WMH, Paris TR, Hardy B (eds) Adoption and diffusion of modernrice varieties in Bangladesh and eastern India. IRRI, Philippines, pp 31–44

Bahu RR, Mishra P (2001) Farm ponds in Coastal Saline Zone of West Bengal: a techno-economic feasibility study.J Agric Eng 38:17–22

Basu D, Banerjee S, Goswami R (2009) Vegetable cultivation on ail (bund): a successful farmer-led technology transfer.Acta Hort 809:155–160

http://www.seamless-ip.org/Reports/Report_53_PD4.4.3.pdf


Bhattacharyya R (2008) Crop Diversification: A Search For An Alternative Income Of The Farmers In The State Of WestBengal In India. Available via http://kastoria.teikoz.gr/icoae2/wordpress/wp-content/uploads/articles/2011/10/012-2008.pdf Accessed 10 May 2013

Bhattacharjee S, Saravanan R (2012) Poverty In Natural Prosperity: Can Agriculture Bring The Renaissance In North-EastIndia? In: Kashyap SK, Pathak A, Papnai G (eds) Saving Humanity Saving Humanity, Swami Vivekananda Perspective.Vivekanand Swadhyay Mandal. GBPUAT, Pantnagar, India, pp 260–274

Bidogeza JC, Berentsen PBM, de Graaff J, Lansink AGJMO (2009) A typology of farm households for the UmutaraProvince in Rwanda. Food Sec 1:321–335

Birner R, Anderson JR (2007) How To Make Agricultural Extension Demand-Driven? The case of india’s agriculturalextension, Discussion Paper 729. International Food Policy Research Institute, Washington DC

Birner R, Davis K, Pender J, Nkonya E, Anandajayasekeram P, Ekboir J, Mbabu A, Spielman D, Horna D, Benin S,Kisamba-Mugerwa W (2009) From best practice to best fit: a framework for designing and analyzing pluralisticagricultural advisory services worldwide. J Agric Edn Ext 15:341–355

Bozeman B (2000) Technology transfer and public policy: a review of research and theory. Res Policy 29:627–655Briggeman BC, Gray AW, Morehart MJ, Baker TG, Wilson CA (2007) A new U.S. farm household typology: Implications

for agricultural policy. Rev Agric Econ 29:765–782Chambers R, Pacey A, Thrupp LA (1989) Farmer First: Farmer Innovation And Agricultural Research. Intermediate

Technology Publications, LondonConroy C, Sutherland A (2004) Participatory technology development with resource-poor farmers: maximising impact

through the use of recommendation domains, AgREN Network Paper no. 133. ODI, LondonDe UK (2002) Economics Of Crop Diversification. Akansha Pub, House, New DelhiDe UK (2003) Diversification of crop in West Bengal: a spatio-temporal analysis. Artha Vijnana 42:170–182Deb UK, Rao GN, Rao YM, Slater R (2002) Diversification And Livelihood Options: A Study Of Two Villages In Andhra

Pradesh, India. Department for International Development (DFID), United KingdomDhiman SC, Thambi DS (2009) Ground water management in coastal areas. Bhu-Jal 24:69–73Ding C, He X (2004) K-Means Clustering Via Principal Component Analysis. In: Proceedings Of The Twenty-First

International Conference On Machine Learning. Banff, Canada. Available via http://ranger.uta.edu/~chqding/papers/KmeansPCA1.pdf. Accessed 16 Aug 2013

Dixon J, Gulliver A, Gibbon D (2001) Farming Systems and Poverty: Improving Farmers’ Livelihoods In A Changing World. FAO, RomeEllis F (1993) Peasant Economics: Farm Households and Agrarian Development, 2nd edn. Cambridge University Press,

CambridgeEmtage N, Suh J (2005) Variations in socioeconomic characteristics, farming assets and livelihood systems of Leyte rural

households. Ann Trop Res 27:35–54Engel PGH (1995) Facilitating Innovation: An Action-Oriented Approach and Participatory Methodology To Improve

Innovative Social Practice In Agriculture. Wageningen Agricultural University, DissertationFafchamps M (1992) Cash crop production, food price volatility, and rural market integration in the third world. Am J

Agric Econ 74:90–99Fernandez-Cornejo J, Hendricks C, Mishra A (2005) Technology adoption and off-farm household income: the case of

herbicide-tolerant soybeans. J Agric Appl Econ 37:549–63Field A (2005) Discovering Statistics Using Spss For Windows, 2nd edn. Sage, LondonFliegel F, Kivlin JE, Sekon GS (1968) A cross-national comparison of farmers’ perceptions of innovations as related to

adoption behavior. Rural Sociol 33:437–449Folmer H, Dutta S, Oud H (2010) Determinants of rural industrial entrepreneurship of farmers in West Bengal: a

structural equations approach. Int Region Sc Rev 33(4):367–396Frankenberger TR, Dewalt BR, McArthur HJ, Hudgens RE, Mitawa G, Rerkasem K, Finan T, Butler Flora C, Young N (1989)

Identification of results of farming systems research and extension activities: a synthesis. FSRE Newsl 1:8–10Gajbhiye KS, Mandal C (2008) Agro–Ecological Zones, Their Soil Resource and Cropping Systems. Available via

http://www.agricoop.nic.in/Farm%20Mech%20PDF/05024-01.pdf. Accessed 29 Nov 2012Ganpat W, Bekele I (2001) Looking For The Trees In The Forest: Farm Typology As A Useful Tool In Defining Targets For

Extension. In: Lindner JR (ed) Proceedings of the 17th Annual Conference of the Association for InternationalAgricultural and Extension Education. Baton Rouge, Louisiana. Available via http://agris.fao.org/agris-search/search.do?f=2002/TT/TT02001.xml;TT2002001006. Accessed 22 February 2014

Gebauer RH (1987) Socio-economic classification of farm household - conceptual, methodical and empirical consider-ations. Eur Rev Agric Econ 14:261–283

Ghosh BK, Kuri PK (2005) Changes in cropping pattern in West Bengal during 1970-71 to 2000-2001. IASSI Quart 24:39–56Glendenning CJ, Babu S, Asenso-Okyere K (2010) Review Of Agricultural Extension In India. Are Farmers Information

Needs Being Met? IFPRI Discussion Paper 01048. IFPRI, Washington DCGómez-Limón JA, Berbel J, Arriaza M (2007) MCDM Farm System Analysis For Public Management Of Irrigated Agriculture.

In: Weintraub A, Bjorndal T, Epstein R, Romero YC (eds) Handbook On Operations Research In Natural Resources InInternational Series In Operations Research And Management Science. Páginas, Kluwer, New York, pp 93–114

Gorsuch RL (1983) Factor Analysis. Erlbaum, Hillsdale, NJGoswami R (2007) Understanding Farmer-To-Farmer Communication Within The Sustainable Rural Livelihood Framework.

Dissertation, Bidhan Chandra Krishi ViswavidyalayaGoswami R, Biswas MS, Basu D (2012) Validation of participatory farming situation identification: a case of rainfed rice

cultivation in selected area of West Bengal, India. Ind J Tradition Knowl 11:471–479Go WB (2005) District Statistical Handbook, South 24 Parganas. Bureau of Applied Economics & Statistics, Government

of West Bengal, IndiaGuto SN, Pypers P, Vanlauwe B, de Ridder N, Giller KE (2010) Socio-ecological niches for minimum tillage and crop-residue

retention in continuous maize cropping systems in smallholder farms of Central Kenya. Agron J 104:188–198Halder JC, Das P (2012) Present status and futuristic view of horticulture in West Bengal. Geo-Analyst 2(1), Available via

http://gswb.in/wp-content/uploads/2012/08/v2n1jully2012_1.pdf. Accessed 10 May 2013

http://kastoria.teikoz.gr/icoae2/wordpress/wp-content/uploads/articles/2011/10/012-2008.pdf

http://kastoria.teikoz.gr/icoae2/wordpress/wp-content/uploads/articles/2011/10/012-2008.pdf

http://ranger.uta.edu/~chqding/papers/KmeansPCA1.pdf

http://ranger.uta.edu/~chqding/papers/KmeansPCA1.pdf

http://www.agricoop.nic.in/Farm%20Mech%20PDF/05024-01.pdf

http://agris.fao.org/agris-search/search.do?f=2002/TT/TT02001.xml;TT2002001006

http://agris.fao.org/agris-search/search.do?f=2002/TT/TT02001.xml;TT2002001006

http://gswb.in/wp-content/uploads/2012/08/v2n1jully2012_1.pdf


Hall AJ, Yoganand B (2004) New Institutional Arrangements In Agricultural Research And Development In Africa:Concepts And Case Studies. In: Hall AJ, Sulaiman RV, Raina RS, Shambu Prasad C, Naik GC, Clark NG (eds)Innovations In Innovation: Reflections On Partnership, Institutions And Learning. International Crops ResearchInstitute for the Semi-Arid Tropics/National Centre for Agricultural Economics and Policy Research, AndhraPradesh/New Delhi, pp 105–134

Hardiman RT, Lacey R, Yang MY (1990) Use of cluster analysis for identification and classification of farming systems inQingyang County, Central North China. Agric Syst 33:115–125

Harrington LH, Tripp R (1984) Recommendation domains: a framework for on-farm research. CIMMYT EconomicsWorking Paper 02/84. Series: CIMMYT Economics Working Paper, CIMMYT, Mexico. Available via http://repository.cimmyt.org/xmlui/bitstream/handle/10883/1001/7013.pdf?sequence=1. Accessed 17 September 2013

Harris RJ (2001) A Primer Of Multivariate Statistics, 2nd edn. Academic, New YorkHayami Y, Ruttan VW (1981) Agricultural Development: An International Perspective. Johns Hopkins University Press,

Baltimore, USAHazell PB (2005) Is there a future for small farms? Agric Econ 32:93–101Jolliffe IT (2002) Principal Component Analysis. Springer Verlag, New YorkJolly CM (1988) The use of action variables in determining recommendation domains: senegalese farmers for research

and Agric Admin Ext 30:253–267Kaiser HF (1970) A second generation Little-Jiffy. Psychomelrika 35:401–415Kaufman L, Rousseeuw PJ (1990) Finding groups in data: an introduction to cluster analysis. Wiley, New YorkKhanna SS (1989) The Agro-Climatic Approach. In: Survey of Indian Agriculture. The Hindu, Madras, India, pp 28–35Kherallah M, Kirsten JF (2002) The New Institutional Economics: applications for agricultural policy research in

developing countries. Agrekon 41:111–134Kobrich C, Rehman T, Khan M (2003) Typification of farming systems for constructing representative farm models: two

illustrations of the application of multi-variate analyses in Chile and Pakistan. Agric Syst 76:141–157Kostrowicki J (1977) Agricultural typology concept and method. Agric Syst 2:33–45Kotler P (1986) Principles of Marketing, 3rd edn. Prentice-Hall, Englewood Cliffs, NJKydd J (2002) Agriculture And Rural Livelihoods: Is Globalisation Opening Or Blocking Paths Out Of Rural Poverty? Agricultural

Research and Extension Network (AgREN) paper no.122. Overseas Development Institute (ODI), London, UKLal P, Lim-Applegate H, Scoccimarro M (2001) The adaptive decision making process as a tool for integrated natural

resource management: focus, attitudes and approach. Conserv Ecol 5:11Leeson JY, Sheard JW, Thomas AG (1999) Multivariate classification of farming systems for use in integrated pest

management studies. Can J Plant Sci 79:647–654Mahapatra AK, Mitchell CP (2001) Classifying tree planters and non planters in a subsistence farming system using a

discriminant analytical approach. Agroforest Syst 52:41–52Malik DP, Singh IJ (2002) Crop diversification – an economic analysis. Ind J Agric Res 36:61–64Mehta R (2009) Rural Livelihood Diversification and Its Measurement Issues: Focus India. Wye City Group on Statistics

On Rural Development And Agriculture Household Income, Second Meeting, at FAO HQ, Italy, Rome, 11-12 June2009. Available via www.fao.org/fileadmin/templates/ess/pages/rural/wye_city_group/2009/Paper_1_a1_Mehta-Wye-Grp-RurDiv-Final.doc Accessed 21 Jul 2012

Mohammad A (1978) Situation Of Agriculture, Food, And Nutrition In Rural India. Concept, New DelhiMwijage NA, de Ridder N, Baijukya F, Pacini C, Giller EK (2009) Exploring the variability among smallholder farms in the

banana-based farming systems in Bukoba district, Northwest Tanzania. Afr J Agric Res 4:1410–1426NABCOM (2009) State Agricultural Plan of Government of West Bengal. NABARD Consultancy Service Private Limited.

Kolkata, India. Available via http://rkvy.nic.in/sap/wb/wb.pdf. Accessed 9 May 2013Nabi R (2008) Constraints to the adoption of rice-fish farming by smallholders in Bangladesh: a farming systems

analysis. Aquacult Econ Manage 12:145–153Nehring R, Fernandez-Cornejo J, Banker D (2005) Off-farm labour and the structure of U.S. Agriculture: the case of

Corn/Soybean farms. Appl Econ 10:633–650North DC (1997) Prologue In The Frontiers Of New Institutional Economics. In: Drobak JN, Nye VCJ (eds) The Frontiers

of The New Institutional Economics. Academic, New YorkOjiem J, Ridder N, Vanlauwe B, Giller KE (2006) Socio-ecological niche: a conceptual framework for integration of

legumes in smallholder farming systems. Int J Agric Sust 4:79–93Orr A, Jere P (1999) Identifying smallholder target groups for IPM in southern Malawi. Int JPest Manage 45:179–187Pal S, Mruthyunjaya J, Joshi PK, Saxena R (2003) Institutional change in Indian agriculture. National Centre for

Agricultural Economics and Policy Research. Indian Council of Agricultural Research, New DelhiPender J, Hazell P (2001) Promoting Sustainable Development In Less-Favored Areas, IFPRI Policy Brief, Focus 4. IFPRI,

New YorkPrein M (2002) Integration of aquaculture into crop–animal systems in Asia. Agric Syst 71:127–146Radulovic V (2005) Are new institutional economics enough? Promoting photovoltaics in India’s agricultural sector. En

Policy 33:1883–1899Reckling M (2011) Characterisation of Bean Farming System Across Farm Types in Northern and Eastern Rwanda.:

Identification Of Potential Niches For Legume Technologies. Dissertation, Wageningen University, The NetherlandsRogers EM (1962) Diffusion of Innovations, 1st edn. The Free Press, New YorkRogers EM (1995) Diffusion of innovation, 4th edn. The Free Press, New YorkRoling N (1988) Extension Science: Information Systems In Agricultural Development. Cambridge University Press, CambridgeRowe EC, van Wijk MT, de Ridder N, Giller KE (2006) Nutrient allocation strategies across a simplified heterogeneous

African smallholder farm. Agric Ecosyst Environ 116:60–71Ruttan VW, Hayami Y (1984) Toward a theory of induced institutional innovation. J Dev Stud 20:203–223Sanjeev MV, Venkatasubramanian V, Singha AK (2010) Farming Dystems Of North East India: Research And

Development Strategies for KVKs. Zonal Project Directorate, Zone – III. Indian Council of Agricultural Research,Umiam, Meghalaya, India

http://repository.cimmyt.org/xmlui/bitstream/handle/10883/1001/7013.pdf?sequence=1

http://repository.cimmyt.org/xmlui/bitstream/handle/10883/1001/7013.pdf?sequence=1

http://www.fao.org/fileadmin/templates/ess/pages/rural/wye_city_group/2009/Paper_1_a1_Mehta-Wye-Grp-RurDiv-Final.doc

http://www.fao.org/fileadmin/templates/ess/pages/rural/wye_city_group/2009/Paper_1_a1_Mehta-Wye-Grp-RurDiv-Final.doc

http://rkvy.nic.in/sap/wb/wb.pdf


Savadogo K, Reardon T, Pietola K (1998) Adoption of improved land-use technologies to increase food security inBurkina Faso: relating animal traction, productivity, and nonfarm income. Agric Syst 58:441–464

Senthilkumar K, Bindraban PS, de Boer W, de Ridder N, Thiyagarajan TM, Giller KE (2009) Characterising rice-basedfarming systems to identify opportunities for adopting water efficient cultivation methods in Tamil Nadu, India.Agric Water Manage 96:1851–1860

Singh S (2002) Contracting out solutions: political economy of contract farming in Indian Punjab. World Dev30:1621–1638

Singh AK, Singh S, Venkatasubramanian V (1993) Relationship between farmers. Socio- personal traits and adoption ofdairy innovations. Charion 22:84–87

Soule MJ (2001) Soil management and the farm typology: do small family farms manage soil and nutrient resourcesdifferently than large family farms? Agric Resour Econ Rev 30:179–188

Sukhatme PV, Sukhatme BV, Sukhatme S, Asok C (1984) Theory Of Sample Surveys With Applications. ISAS, IASRI,New Delhi

Thomas HC (ed) (2006) Trade Reforms And Food Security: Country Case Studies And Synthesis. Food and AgricultureOrganisation of UN, Rome

Timothy WO (1994) Identifying target groups for livestock improvement research: the classification of sedentarylivestock producers in western Niger. Agric Syst 46:227–237

Tittonell P, Vanlauwe B, Leffelaar PA, Shepherd KD, Giller KE (2005) Exploring diversity in soil fertility management ofsmallholder farms in western Kenya. II. Within-farm variability in resource allocation, nutrient flows and soil fertilitystatus. Agric Ecosyst Environ 110:166–184

Tittonell P, Leffelaar PA, Vanlauwe B, van Wijk MT, Giller KE (2006) Exploring diversity of crop and soil managementwithin smallholder African farms: a dynamic model for simulation of N balances and use efficiencies at field scale.Agric Syst 91:71–101

Tittonell P, Vanlauwe B, de Ridder N, Giller KE (2007) Heterogeneity of crop productivity and resource use efficiencywithin smallholder Kenyan farms: soil fertility gradients or management intensity gradients. Agric Syst 94:376–390

Tittonell P, Muriukid A, Shepherde KD, Mugendif D, Kaizzig KC, Okeyoa J, Verchote L, Coee R, Vanlauwea B (2010) Thediversity of rural livelihoods and their influence on soil fertility in agricultural systems of East Africa – A typology ofsmallholder farms. Agric Syst 103:83–97

Usai MG, Casu S, Molle G, Decandia M, Ligios S, Carta A (2006) Using cluster analysis to characterize the goat farmingsystem in Sardinia. Livest Sci 104:63–76

USDA, ERS (2000) ERS Farm Typology For A Diverse Agricultural Sector, Agriculture Information Bulletin Number 759.USDA, US. Accessed via http://ageconsearch.umn.edu/bitstream/33657/1/ai000759.pdf. Accessed 12 August 2012

Vanclay JK (2005) Using a typology of tree-growers to guide forestry extension. Ann Trop Res 27:97–103Williamson OE (1990) A comparison of alternative approaches to economic organization. J InstTheor Econ 146:61–71Zingore S, Murwiraa HK, Delvea RJ, Gillerb KE (2007) Soil type, management history and current resource allocation:

Three dimensions regulating variability in crop productivity on African smallholder farms. Field Crops Res101(3):296–305

doi:10.1186/s40100-014-0005-2Cite this article as: Goswami et al.: Farm types and their economic characterization in complex agro-ecosystems forinformed extension intervention: study from coastal West Bengal, India. Agricultural and Food Economics 2014 2:5.

Submit your manuscript to a journal and benefi t from:

7 Convenient online submission

7 Rigorous peer review

7 Immediate publication on acceptance

7 Open access: articles freely available online

7 High visibility within the fi eld

7 Retaining the copyright to your article

Submit your next manuscript at 7 springeropen.com

http://ageconsearch.umn.edu/bitstream/33657/1/ai000759.pdf

research open access farm types and their economic ... · identification and characterization of...

Documents